HIE-ISOLDE: THE SUPERCONDUCTING RIB LINAC AT CERN

M. Pasini on behalf of the HIE-ISOLDE design teamCERN, Geneva, CH and Instituut voor Kern- en Stralingsfysica, K.U.Leuven, Leuven, BE

Abstract

In the frame of the upgrade of the ISOLDE facility atCERN, a R&D program on superconducting linac for Ra-dioactive Ion Beams (RIBs) has started in 2008. The linacwill be based on superconducting Quarter Wave Resonators(QWRs) and will make use of high field SC solenoids forthe beam focusing. The sputtering technology has beenchosen as the baseline technique for the cavity manufac-turing and prototype and sputtering tests are in advancedstate. A status report on the SC activities will be presented.

INTRODUCTION

In the present REX-ISOLDE facility the RIBs are accel-erated to high energies with a compact Normal Conducting(NC) linac, making use of a special low energy preparatoryscheme where the ion charge state is boosted so that themaximum mass to charge ratio is always 2.5 < A/q < 4.5.This scheme consists of a Penning trap (REXTRAP), acharge breeder (REXEBIS) and an achromatic A/q sepa-rator of the Nier spectrometer type. The NC acceleratoris designed with an accelerating voltage for a correspond-ing maximum A/q of 4.5 and it delivers a final energy of3 MeV/u for A/q < 3.5 and 2.8 for A/q < 4.5. Aftercharge breeding, the first acceleration stage is provided bya 101.28 MHz 4-rod Radio Frequency Quadrupole (RFQ)which takes the beam from an energy of 5 keV/u up to300 keV/u. The beam is then re-bunched into the first101.28MHz interdigital drift tube (IH) structure which in-creases the energy to 1.2 MeV/u. Three split ring cavitiesare used to give further acceleration to 2.2 MeV/u and fi-nally a 202.58 MHz 9-gap IH cavity is used to boost andto vary the energy between 2 < E < 3 MeV/u. Fig. 1illustrates the scheme of the present linac.

Figure 1: REX-ISOLDE present scheme.

The operation of the REX complex has started in 2002and it sports now the largest variety of accelerated radioac-tive ion beams available worldwide [1].

In order to respond to the ever increasing demand forbeams of higher energy, higher intensity and better quality(purity and emittance-wise) [2], a new upgrade programhas been launched under the name of HIE-ISOLDE.

The higher energy requirement is achieved by means of amodular superconducting linac based on QWRs which willbe installed downstream the normal conducting linac. Inthe short term the new accelerator modules will boost theenergy up to 5.5, 8 and 10 MeV/u, while in the longer term,part of present normal conducting linac will be replaced bynew superconducting cavities in order to allow the full en-ergy variability between 1.2 and 10 MeV/u [3]. The stagingof the installation allows to minimize the down time of thephysics program and to profit of the periodic winter shut-down period to commission the machine. Fig 2 shows apossible scheme for the machine installation.

Figure 2: HIE-ISOLDE installation staging.

Concerning the higher beam intensity, ISOLDE willprofit from the ongoing upgrade of the proton injectorschain at CERN [4] which will allow the beam power on tar-get to be doubled and for which new target stations, targetsand their associate handling system will need new develop-ment. Moreover, an upgrade of the REX trap and chargebreeder is planned to cope with the increased intensity.

Finally in order to improve the quality of the beam a newmass separator with higher resolution is under study andnew targets and ion sources are under development.

Because of the limited resources available the priorityof the HIE-ISOLDE project has been given to the designand construction of the SC linac. In particular the R&Deffort has focused on the development of the high β cav-ity (β = 10.3%), for which it has been decided to adoptthe Nb sputtered on Cu substrate technology. Other R&Dactivities are related to the beam dynamics studies whichseek to define a very compact accelerating lattice and con-sequently the shortest possible machine, a design of com-pact SC solenoids with limited fringe fields, and the studyof the cryomodule concept.

CAVITIES DESIGN

The superconducting linac is designed to deliver an ef-fective accelerating voltage of at least 39.6 MV with an

FROBAU03 Proceedings of SRF2009, Berlin, Germany

02 Future projects

924

average synchronous phase φs of -20 deg. This is the min-imum voltage required in order to achieve a final energy ofat least 10 MeV/u with A/q = 4.5. Because of the steepvariation of the ions velocity, at least two cavity geome-tries are required in order to have an efficient accelerationthroughout the whole energy range. A total number of 32cavities are needed to provide the full acceleration voltage.The geometries chosen corresponding to low (β0 = 6.3%)and high (β0 = 10.3%) “β” cavities maintain the funda-mental beam frequency of 101.28 MHz and their designparameters are given in Table 1. The design acceleratinggradient aims at reaching 6 MV/m with a power consump-tion of 7 W per low β cavity and 10 W per high β cavity.

Table 1: Cavity design parameters

Cavity Low β highβ

No. of Cells 2 2f (MHz) 101.28 101.28β0 (%) 6.3 10.3Design gradient Eacc(MV/m) 6 6Active length (mm) 195 300Inner conductor diameter (mm) 50 90Mechanical length (mm) 215 320Gap length (mm) 50 85Beam aperture diameter (mm) 20 20U/Eacc

2 (mJ/(MV/m)2 73 207Epk/Eacc 5.4 5.6

Hpk/Eacc (Oe/MV/m) 80 96

Rsh/Q (Ω) 564 554Γ = Rs · Q0 (Ω) 23 30.34Q0 for 6MV/m at 7W 3.2 · 108 7 · 108

TTF max 0.85 0.9No. of cavities 12 20

Coupler

An adjustable power coupler has been designed with alarge dynamic range. In fact the Qext can be adjusted be-tween 104 up to 109 so that it is possible to reach criticalcoupling at room temperature and a moderate undercou-pling when the cavity is superconducting. This feature al-lows to perform some preliminary room temperature con-ditioning that should help in cleaning the multipacting bar-rier at low field level and allow to fine tune the couplingneeded for the field and phase stabilisation loops. From themechanical point of view the sliding mechanism concept isdust free as the contact points are reduced to the minimum(see Fig. 3). This should avoid any seizure of the movingparts and guarantees the functionality of the system. A fullelectromagnetic design of the coupler line has been per-formed and reported in [5]. The mechanical design is nowfrozen and construction is expected to be completed by endof November.

Figure 3: Coupler schematics.

Tuner

The tuning system chosen for the HIE-ISOLDE cavi-ties takes advantage from the experience developed at TRI-UMF [6]. An oil-can shaped diaphragm of CuBe has beenhydroformed with a pressure up to 120 bar. All radialslot necessary for the elongation and contraction of the di-aphragm are performed with a laser beam. The same platecan be mounted directly on the low β cavity or welded toa flange in the case of the high β cavity. The actuator isdesigned to have no backlash. A coarse frequency tuningrange of 220 kHz is expected. Such large tuning range (forthis type of cavity) allows to maintain the tolerance on thegeometry to a value in the order of tenth of mm. The firstprototype of the tuner is available and the internal surfaceshould be sputtered by the end of 2009.

Figure 4: 3D view of the tuner with the lever actuator.

BEAM DYNAMICS

The main goal of the beam dynamics study is to definean optics and acceleration scheme that minimises the ma-chine length and at the same time maintains the emittanceof the beam. The resulting lattice has been discussed in [7]and latest results are reported in [8]. In the high β sec-tion the lattice consists in a short drift length outside the

Proceedings of SRF2009, Berlin, Germany FROBAU03

02 Future projects

925

cryomodule where all the beam instrumentation and smallcorrector magnets will be installed, followed by two super-conducting cavities, a single SC solenoid and finally threemore SC cavities (see Fig. 5).

Figure 5: Schematics of the high energy cryomodule.

A full three dimensional integration of the motion equa-tion routine has been written in order to study the motion ofthe single particle inside the high β cavity. The electromag-netic cavity fields used in the tracking code were calculatedusing the MWS code [9]. The multi-particle beam dynamicssimulations are performed by implementing the fields mapin the TRACK code [10].

An intrinsic characteristic of the QWR is the asymmetryof the electromagnetic field in the beam region. There is infact a net magnetic component which steers the beam de-pending on the accelerating phase. The radial electric fieldis also not symmetric and the compensation scheme pro-posed by Ostroumov [11] would lead to a loss of apertureof roughly 30% in case of a circular aperture. It was thendecided to modify the shape on the beam port from circu-lar to a race-track one, and the loss of aperture due to thecompensation scheme is now only 1%. Fig. 6 shows theeffect of the compensation scheme. The net diverging kickgiven to the beam is reduced to less than 0.1mrad only forthe first cavity and as soon as the beam picks more velocitythe kick becomes rapidly negligible.

Fig. 7 shows the result of the beam dynamics study ofthe high energy section in case of a beam with A/q=4.5 andwith a transversed matched beam coming from the presentNC machine.

The beam dynamics study in now focused on the errorstudy, considering longitudinal fast error (RF jitter) andstatic error as well as static transverse misalignment. Itis found that the longitudinal emittance in the high energysection of the machine increases by 20% if the RF jitter islimited at 0.5% in amplitude and at 0.5 deg in phase for abeam with A/q=4.5. In the latest study a high order analy-sis of the defocusing term has been performed and resultsshows that even in the case of solenoid focusing with asym-metric beams the transverse emittance is still kept undercontrol [12].

0.08 0.09 0.1 0.11 0.12 0.13 0.14 0.15

−0.45

−0.4

−0.35

−0.3

−0.25

−0.2

−0.15

−0.1

−0.05

0

Injection Velocity, β

Ver

tical

Bea

m S

teer

ing

Kic

k, Δ

y’ (

mra

d)

2.6 mm OffsetOn Axis

Figure 6: Divergence of the beam in case of a compensated(red) and not compensated (blue) case for a synchronousphase of -20 deg as a function of the incoming beam veloc-ity.

Em

it.T

G’th

(%)

012345

012345

Em

it.L,

RM

S G

’th (%

)

02468

Em

it.L,

99.5

% G

’th (%

)

020406080

E (M

eV/u

)

0.0 1.2 2.4 3.6 4.8 6.0 7.2 8.4 9.6Distance (m)

02468

Envelope

99.5%

99.5%

RMS

RMS

RMS

Bea

m E

nv. (

mm

)

012345

012345

Figure 7: Beam dynamic of A/q=4.5 in the case of the highenergy section only.

SC Solenoid

The employment of the SC solenoids as unique focusingelements allows to increase the transverse acceptance ofthe beam with respect to the un-matched beam conditionand allows a great reduction of the tuning knobs, hencemaking the tuning and operation of the machine easier. Themagnetic and mechanical specifications of the solenoid aresummarized in Table 2.

Table 2: Solenoid specifications

Magnetic Integral∫

B2dz 16.2 T2 mMechanical Length 0.4 mFringe field at 23 cm from center < 0.2 GInner Diameter 30 mm

A 2D electromagnetic study of the solenoid has been per-formed: the specifications are met with 360A Nb3Sn coilswhich produces a peak field in the center of the solenoid of120kG. The return yoke is made by high quality iron. The

FROBAU03 Proceedings of SRF2009, Berlin, Germany

02 Future projects

926

fringe fields level are at the level of 50G when the solenoidis powered at the maximum field. This value is largely safewhen the QWRs next to the solenoid are superconducting.Fig. 2 show a schematic of the mechanical assembly of thesolenoid.

Figure 8: Solenoid schematics.

A comparison of the parameters between Nb-Ti andNb3-Sn solenoid giving the same field has been performedand results are listed in Table 3. A model for checking thetools and assembly procedures of the coils is in preparation.

Table 3: Parameters comparison between Nb-Ti and Nb3-Sn solenoid

Parameters Nb-Ti Nb3-Sn

Number of turns # 16500 4500Wire length (m) 5200 740Working point (%) 93 55Operating current (A) 100 360Inductance (mH) 3500 142Stored energy (kJ) 17.5 9.5Radial force (kN/rad) 250 145Coil weight (kg) 36 5.2Inner diameter (mm) 34 34Outer diameter (mm) 166 70Solenoid length (mm) 250 250

HIGH β CAVITY MANUFACTURINGThe construction of a high β cavity prototype started

in the middle of 2008 and the copper body which makesthe substrate for the Nb sputtering was completed in April2009. During the same period a new sputtering chamberwas constructed and is now under commissioning. Thecavity manufacturing process is reported in [13] and nowthe critical fabrication steps have been identified. In de-tails the long longitudinal full penetration electron beamweldings have shown the creation of some porousness thatcould harm the Nb deposit. This defect could be removed

by passing the electron beam in smoothing mode from theinside of the cavity - this operation would require the useof a special electron beam welding machine with a minia-turized head - but it was not felt as a immediate action totake. Moreover after the chemical etching all the micro-difects seems to be disappeared and hence not harmful forthe subsequent Nb sputtering. The most critical and alsotime-consuming step was the direct extrusion of the beamport. This step requires high precision in mounting andremounting the different tools and particular care in han-dling the cavity since for the production of the beam ports,the cavity needs to undergo to eight heat treatments whichlocally soften the copper. Finally the construction of theprototype suffered a couple of accidents which are relatedto human errors. For this reason a strong QA procedureneeds to be in place for the series production. The cavityhas been checked also with a series of RF measurement,especially the variation of the resonance frequency duringthe several manufacturing steps and all the measurementsare in line with the prediction. The mechanical fabricationof the copper cavity has been completed and the chemicalpolishing has been performed. Details about the differentprocedures for the surface treatment are reported in [14]Fig. 9 shows the cavity after the final low water pressurerinsing subsequent to the chemical polishing.

Figure 9: Cavity copper substrate towards final machiningsteps.

In parallel to the copper cavity a dummy cavity made instainless steel was built. The purpose of this dummy cav-ity is to use it as samples holder for the characterization ofthe plasma inside the chamber, to test the assembly proce-dure of the sputtering chamber itself and to serve also as atraining tool for the handling operation during the chemicaltreatment.

The sputtering chamber is now operational and the firstplasma has been produced. The vacuum level of the cham-ber after baking reached the 10−9 mbar level and a proce-dure is now in place to routinely obtain this value. Thisis of course of great importance as the vacuum level is di-rectly linked to the quality of the Nb films produced. Allsubsystems have been checked and a characterization of theplasma and hence of the Nb deposit quality is ongoing as a

Proceedings of SRF2009, Berlin, Germany FROBAU03

02 Future projects

927

function of the different gas pressure level, cathode voltageand bias voltage. First Nb deposit on the copper cavity isexpected towards the end of October. Fig. 10 shows on theleft a moment during the installation of the dummy cavityinside the sputtering chamber and on the right the sputter-ing chamber closed.

Figure 10: On the left the dummy cavity during mountingfor the first sputtering. On the right the sputtering chamberclosed

CRYOMODULES

An initial study of the cryomodule concept was reportedin [15]. Specifically, in that report the choice for a singlevacuum system with an active thermal shield was shown tobest fit the requirements of the new machine. In the lastmonths, work has continued at a level of a concept studyand a series of major choices have been taken concern-ing the dimensions, accessibility, assembly procedure andmaintenance of the cryomodule, the holding system for thecavities and for the solenoid, and the alignment system.

One of the general consideration that set the pathwayto the concept of the HIE-ISOLDE cryomodule was themaintenance service. In order to minimize cost and down-time of repairs it is clear that the maintenance should takeplace at CERN and that at this point one needs to con-sider the available infrastructure such as clean rooms. Allthe clean rooms at CERN are quite limited in height butthey are quite large and long. As a consequence the dis-mounting of the cryomodule can occur only if the access isfrom the lateral side (see Fig. 11). Starting from this con-sideration a study concerning the stability of a mechanicalstructure with two openings on the side has been performedand found that this structure is feasible. The mounting andpre-alignemt of the cavities will be externally in the sup-port girder which, with the help of a special forklift can bepositioned inside the cryomodule and hooked to the sup-port stems. Given the recent result in the error study it hasbeen decided to have a separate frame for the support of thesolenoid which can be adjusted from outside even when thesystem is cold.

Concerning the alignment system, it is under study thepossibility to use a BCAM system [16]. Such a system

consists of a calibrated laser source that can track changesin the position with an accuracy in the order of few tens ofμm with a rather large span in distances. The advantage ofsuch system is that the active part of the alignement systemis kept outside the vacuum envelope and that it is possibleto perform a continuous survey so that in case of accidentone can retrace the position of the different elements justbefore the accident. A small test set-up to test the windowperturbation of the measurement is being prepared.

Figure 11: Cryomodule schematics.

INFRASTRUCTURES

A study covering all the issues of the infrastructure andthe integration of the machine inside the experimental hallis ongoing. This has permitted to identify the suitable areafor the construction of the building for the compressor forthe liquid He system, the location of the He liquefier withthe associated control room and also the other subsystemlike the new ventilation unit, the new electricity distribu-tion panels and the position of the different racks. Thisaspect of the project cannot be left behind as the locationof such infrastructures determines the free space left forthe accelerator, for the high energy beam transfer line andfor the physics instrumentation. Fig. 12 shows the resultof the study. Around the linac a semi-permanent tunnel-like shielding made of concrete blocks will be installed allalong the machine length. This is strictly required fromthe radioprotection point of view as the whole area will re-main accessible during the physics runs. The LHe liquefierwill be installed at the end of the machine in a separate lightconstruction building. In this way it is possible to minimizethe length of the LHe distribution system and to maintainthe geometry of the line as simple as possible which is acondition for a easier and stable operation of the cryogenicsystem.

SUMMARY

The HIE-ISOLDE proposal has been submitted to theCERN management which has recognized the scientific

FROBAU03 Proceedings of SRF2009, Berlin, Germany

02 Future projects

928

Figure 12: ISOLDE facility planimetric view. In the bottom left part one can identify the compressor building, while onthe centre right there is the He liquefier. This configuration minimizes the length of the LHe distribution line, and set thecondition for the optimum operation of the cryogenic system.

value of the project. The R&D cavity will continue withthe production of 5 more cavities and with the constructionof a single cavity test cryostat. RF tests will be initiallyperformed at TRIUMF and the validation of the cavity con-struction procedure will follow soon after.

ACKNOWLEDGMENTS

The author wish to acknowledge the talented and verysupportive design group at CERN. I wish also to thank R.Laxdal and P. Ostroumov for their suggestion and com-ments of the machine design. Finally I wish to thank E.Palmieri for the strong support and help in the developmentof the sputtering technology for the QWRs.

REFERENCES[1] D. Voulot, et al., Radioactive beams at REX-ISOLDE:

Present status ..., Nucl. Instr. and Meth. B (2008),doi:10.1016/j.nimb.2008.05.129

[2] HIE-ISOLDE: the scientific opportunities, CERN Report2007-08.

[3] M. Pasini et al., “A SC upgrade for the REX-ISOLDE accel-erator at CERN”, Linac’08, Victoria, Canada, Sept. 2008.

[4] http://linac4.web.cern.ch/linac4/.

[5] A. D’Elia et al., “Design and Characterization of the PowerCoupler Line for HIE-ISOLDE High Beta Cavity”, TheseProc.

[6] T. Ries et al., “A mechanical Tuner for the ISAC-II QuarterWave Superconducting Cavities”, PAC’03, Portland, USA,May 2003.

[7] M. Pasini et al., “Beam Dynamics Studies for the SCREX-ISOLDE linac at CERN”, Linac’08, Victoria, Canada, Sept.2008.

[8] M. A. Fraser et al., “Beam Dynamics Studies for the HIE-ISOLDE Linac at CERN”, PAC’09, Vancouver, Canada,May 2009.

[9] http://www.cst.com

[10] P. Ostroumov at al.,http://www.phy.anl.gov/atlas/TRACK/

[11] P. Ostroumov and K. Shepard, Phys. Rev. ST. Accel. Beams11, 030101(2001).

[12] M. Fraser at al., “Compensation of transverse asymmetri inthe high-beta quarter-wave resonator of the HIE-LINAC”,these Proc.

[13] S. Calatroni et al., “The HIE-ISOLDE SuperconductingCavities: Mechanical Design and Fabrication”, these Proc.

[14] G. Lanza et al.,“The HIE-ISOLDE Superconducting Cavi-ties: Surface Treatment and Niobium Thin Film Coating”,these Proc.

[15] V. Parma et al., “Concept Design Studies of the REX-ISOLDE cryomodules at CERN”, Linac’08, victoria,Canada, Sept. 2008.

[16] http://alignment.hep.brandeis.edu/Devices/BCAM/

Proceedings of SRF2009, Berlin, Germany FROBAU03

02 Future projects

929